Path planning method and device for adaptive sampling rope-driven parallel transfer robot

Abstract

The invention provides a path planning method and device for a self-adaptive sampling rope-driven parallel transfer robot, and relates to the field of path planning. The search space of the robot is determined by acquiring the starting point, the target terminal point, the relevant information of the robot and the like; when the initial state of the robot is feasible, respectively creating random trees by taking a starting point and a target end point as root nodes; respectively carrying out search sampling in respective search spaces by adopting a bidirectional self-adaptive growth adjustment strategy, and finding a next sub-node in the feasible path; traversing each adjacent node of the current child node, and selecting an optimal father node; continuously and alternately expanding new nodes of the two random trees and trying to connect until the construction of the random trees is completed, and generating a heuristic trajectory; according to the heuristic trajectory, controlling the robot to move to the next child node along the trajectory; and updating the state and environment information of the robot, and re-sampling and expanding new nodes until the robot reaches a target terminal point. According to the invention, an efficient, safe and feasible path can be generated.

Description

Technical Field

[0001] The present invention relates to the field of path planning and artificial intelligence technology, and in particular to a path planning method and device for an adaptive sampling rope-driven parallel handling robot. Background Art

[0002] With the rapid development of industrial automation and intelligent manufacturing, rope-driven parallel handling robots are becoming an effective alternative to traditional lifting equipment due to their lightweight, flexible, and efficient design. These robots have broad application prospects in areas such as rocket and satellite manufacturing, ship maintenance, and automated warehousing.

[0003] However, the path planning problem of rope-driven parallel transport robots remains one of the core challenges in their technological development. The goal of path planning is to generate a safe and feasible path from the starting point to the target point for the robot in a complex working environment while avoiding collisions with obstacles. Existing path planning methods are usually based on random tree expansion algorithms (such as RRT) for sampling and searching, but these methods have many shortcomings in practical applications. For example, traditional algorithms perform uniform random sampling in free space. Although it can cover a large search range, it is inefficient when approaching the target point or in areas with dense obstacles, and it is difficult to quickly find the optimal path. In addition, the existing methods have limited perception of environmental changes and may not be able to adjust the path in real time in a dynamic environment, resulting in inaccurate or even invalid planning results.

[0004] A more prominent problem is that existing path planning methods are inadequate in optimizing path length and obstacle avoidance, resulting in lengthy or inefficient paths that waste time and space. Furthermore, for devices with a unique structure like a rope-driven parallel robot, the kinematic characteristics of the end effector and rope place even higher demands on path planning. For example, ensuring the end effector's trajectory is both safe and efficient, rapidly detecting and avoiding obstacles, and updating the path in real time in a dynamic environment are all pressing technical challenges.

[0005] In view of this, the applicant filed this application after studying the existing technology. Summary of the Invention

[0006] The present invention aims to provide a path planning method and device for an adaptive sampling rope-driven parallel transport robot. The method aims to improve the efficiency and quality of path planning through bidirectional random tree expansion and adaptive sampling strategy, so as to solve the problems of low efficiency, insufficient safety and poor dynamic adaptability of existing rope-driven parallel transport robot path planning methods in complex environments.

[0007] In order to solve the above technical problems, the present invention is implemented through the following technical solutions:

[0008] A path planning method for an adaptive sampling rope-driven parallel handling robot, comprising:

[0009] S1, obtaining the starting point, target end point, environmental information, the initial state of the robot, and the workspace, and determining the robot's search space; wherein the robot's initial position is the same as the starting point; the environmental information includes the current location of obstacles;

[0010] S2, checking the initial state of the robot according to the position of the workspace and the current obstacle, and creating a random tree with the starting point and the target end point as root nodes respectively in a feasible initial state;

[0011] S3, based on the current state of the random tree, a bidirectional adaptive sampling strategy is used to search and sample in each search space to find the next child node in the feasible path;

[0012] S4, traverse each neighbor node within the range of the current child node found in turn, calculate the path cost from the root node of the current expanded random tree to the current child node, and select the neighbor node with the smallest path cost as the best parent node of the current child node;

[0013] S5, after each new node is generated, the new nodes of the two random trees are continuously expanded alternately and the two random trees are tried to be connected until the two random trees have the same node or the distance between the generated new nodes is less than the set threshold. With the root nodes of the two random trees as the starting and end points, the child nodes with the smallest path cost are searched to generate a heuristic trajectory to complete the construction of the random tree;

[0014] S6, according to the generated heuristic trajectory, control the robot to move from the current node along the trajectory to the next child node;

[0015] S7, updates the robot's state and environment information, takes the robot's current state node as the root node of the new random tree, resamples and expands new nodes to search for a better path until the robot reaches the target end point.

[0016] Preferably, the search space of the robot is determined based on the robot's workspace and obstacle space; the robot's workspace is a set of feasible poses of its end effector, and the pose includes the position and orientation of the end effector, and the position is expressed in a Cartesian coordinate system;

[0017] Apply spatial constraints to the robot's workspace. The spatial constraints are expressed as:

[0018]

[0019] x rob(t) ∈x,x∈R d ;

[0020] Among them, t f represents the total time; t represents the moment variable; x rob represents the robot's end effector space; x rob(t) represents the state of the robot at time t; w represents the robot's workspace; x represents the robot's d-dimensional state space; R represents a real number;

[0021] Obstacle space x obs , Indicates the state where the robot collides with an obstacle;

[0022] Free Spacex free , x free =x\x obs A set of states that represent a robot that does not collide with obstacles.

[0023] Preferably, the robot is in a feasible initial state when it meets the following conditions:

[0024] The initial state of the robot satisfies the spatial constraints of the workspace;

[0025] The state of the robot at the current time t belongs to the free space x free The range of x rob(t) ∈x free ;

[0026] Calculate the minimum distance between the robot and the obstacle, which is within the safety threshold range, that is:

[0027]

[0028] Among them, C i (t), O i (t) represents the two points closest to the robot's end effector and the obstacle at time t; L represents the set safety threshold.

[0029] Preferably, when calculating the minimum distance between the robot and the obstacle, the robot's end effector and the obstacle are modeled as convex bodies, and the vertex information of the convex body is extracted to form a corresponding vertex set; and the GJK algorithm is used to calculate the minimum distance between the robot and the obstacle, that is, the Minkowski difference of the two convex body vertex sets is calculated, and the closest point is found in the Minkowski difference, which is the minimum distance between the robot and the obstacle.

[0030] Preferably, the adaptive sampling strategy is specifically:

[0031] If a path to the target destination has been found, sampling is performed within an elliptical area with the current path as the core to optimize the current path; wherein the focus of the ellipse is determined based on the target starting point and the target destination, and the major axis of the ellipse is set based on the length of the current path;

[0032] If no path to the target destination is found, the weight between the target bias strategy and the random sampling strategy is adaptively adjusted according to the node expansion failure rate;

[0033] When the node expansion failure rate is less than the set threshold, the weight of the target bias strategy is increased, and the sampling is expanded in the target direction; otherwise, the weight of the random sampling strategy is increased, and the search range is expanded for sampling; the node expansion failure rate is the ratio of the number of node expansion failures to the total number of expansions;

[0034] The formula for adaptive sampling is:

[0035] P sample =α×P goal +(1-α)×P random ;

[0036] Among them, P sample is the sampling node position; α is the weight of the target bias strategy, 0≤α≤1; P goal is the position of the target direction; P random is a random position in free space;

[0037] A bidirectional adaptive sampling strategy is used to sample a random node x in the robot’s search space. rand After that, get the random node x in the current random tree T1 rand The nearest child node x nearest1 , and along x nearest1 To a random node x rand Expand the direction by one step to generate a new node x new1 ;

[0038] New node x new1 Use its neighbor nodes to optimize the random tree T1 and select the neighbor node with the smallest path cost as x new1 The best parent node of

[0039] Generate a new node x new1 After that, swap the current operation tree, try to connect the two random trees, and find the closest child node x in the random tree T2. nearest2 , along x nearest2 To x new1 Expand one step in the direction to generate a new node x new2 ; New node x new2Use its neighbor nodes to optimize the random tree T2 and select the neighbor node with the smallest path cost as x new2 The best parent node of

[0040] If the new node expansion fails or the connection attempt between two random trees fails, the current random tree structure is retained and the next iteration is started to generate a new random sampling point x rand , and continue to expand random trees T1 and T2 alternately.

[0041] Preferably, when determining whether the new node is the next child node in the feasible path, the determination method is:

[0042] The GJK algorithm is used to detect whether the generated new node is in the robot's workspace. If not, the new node is discarded and resampled and expanded;

[0043] Otherwise, determine whether the minimum distance between the robot and the obstacle under the new node is greater than the safety threshold; if it is, the new node is the next child node;

[0044] Otherwise, further estimate whether there is a collision between the path from the new node's neighbor node to the new node and the obstacle; if there is no collision, it is a safe and feasible path, and the new node is the next child node;

[0045] Otherwise, discard the new node and resample.

[0046] Preferably, when traversing each neighbor node within the domain of the current child node of the current expanded random tree, the domain radius R is set to:

[0047]

[0048] Among them, R η >0 is a pre-constant, d′ represents the problem dimension, and n is the number of nodes.

[0049] Preferably, when the root nodes of the two random trees are used as the starting point and the end point, a heuristic trajectory is generated by searching for the child nodes with the smallest path cost. A total cost function is constructed based on the heuristic function and the path cost, and the optimal branch in the path is selected by calculating the total cost value. The calculation formula of the total cost function is:

[0050] f i =cost(x i )-h i ;

[0051] h i =||x i -x goal ||;

[0052] Among them, x iIndicates the current node, fi i Represents the total cost function of the current node; cost(x i ) represents the path length of the current node, that is, the path from the root node to the current node x i The path cost of h i Represents the current node x i To the target end point x goal expected costs.

[0053] Preferably, it also includes: if no feasible heuristic trajectory is generated, the robot stops on the spot to prevent danger until a new feasible trajectory is generated; when the robot moves from the current node to the next child node, the node and its child nodes with a timestamp earlier than the current node are deleted to optimize the path.

[0054] The present invention also provides a path planning device for an adaptive sampling rope-driven parallel transport robot, comprising:

[0055] An acquisition unit, configured to acquire a starting point, a target end point, environmental information, an initial state of the robot, and a workspace, and determine a search space for the robot; wherein the initial position of the robot is the same as the starting point; and the environmental information includes the current position of obstacles;

[0056] a random tree creation unit, configured to check an initial state of the robot according to the position of the workspace and the current obstacle, and create a random tree with the starting point and the target end point as root nodes respectively in a feasible initial state;

[0057] The bidirectional sampling unit is used to search and sample in each search space according to the current state of the random tree using a bidirectional adaptive sampling strategy to find the next child node in the feasible path;

[0058] The path cost calculation unit is used to traverse each neighbor node within the range of the current child node, calculate the path cost from the root node of the current extended random tree to the current child node, and select the neighbor node with the smallest path cost as the best parent node of the current child node;

[0059] The heuristic trajectory generation unit is used to alternately expand the new nodes of the two random trees and try to connect the two random trees after each new node is generated until the two random trees have the same node or the distance between the generated new nodes is less than the set threshold. With the root nodes of the two random trees as the starting and end points, the unit searches for the child nodes with the smallest path cost to generate a heuristic trajectory and complete the construction of the random trees.

[0060] The control unit is used to control the robot to move from the current node along the trajectory to the next child node according to the generated heuristic trajectory;

[0061] The update unit is used to update the robot's state and environment information, taking the robot's current state node as the root node of the new random tree, resampling and expanding new nodes to search for a better path until the robot reaches the target end point.

[0062] The present invention also provides a path planning device for an adaptive sampling rope-driven parallel transport robot, comprising a processor and a memory, wherein the memory stores a computer program, and the computer program can be executed by the processor to implement a path planning method for an adaptive sampling rope-driven parallel transport robot as described above.

[0063] The present invention also provides a computer-readable storage medium, on which computer-readable instructions are stored. When the computer-readable instructions are executed by a processor of a device where the computer-readable storage medium is located, the path planning method of the adaptive sampling rope-driven parallel transport robot as described above is implemented.

[0064] In summary, compared with the prior art, the present invention has the following beneficial effects:

[0065] First, the present invention significantly improves path planning efficiency through bidirectional random tree expansion and an adaptive sampling strategy, making it particularly suitable for complex working environments. The adaptive sampling strategy dynamically adjusts sampling weights based on the node expansion failure rate, thereby improving planning efficiency when approaching targets or areas with dense obstacles.

[0066] Secondly, the invention incorporates a fast collision detection algorithm to ensure that the generated path will not collide with obstacles, improving path safety. In particular, the sweeping body scanning detection method effectively assesses the potential collision risk with obstacles during robot movement.

[0067] Thirdly, the present invention has strong dynamic adaptability and can update the path in real time to cope with environmental changes.

[0068] Finally, the present invention generates the shortest and safest transport path through an optimization strategy combining cost function and heuristic function, which meets the needs of the rope-driven parallel transport robot in the process of transporting goods. BRIEF DESCRIPTION OF THE DRAWINGS

[0069] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments. It should be understood that the following drawings only illustrate certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other relevant drawings can be obtained based on these drawings without paying any creative work.

[0070] Figure 1This is a flow chart of a path planning method for an adaptive sampling rope-driven parallel handling robot provided in Example 1.

[0071] Figure 2 This is a simplified structural diagram of the rope-driven parallel handling robot provided in Example 1.

[0072] Figure 3 This is a schematic diagram of the relative positions of the robot and the obstacle at time t provided in Example 1.

[0073] Figure 4 This is an example diagram of robot obstacle avoidance path planning with bidirectional adaptive sampling provided in Example 1.

[0074] Figure 5 A schematic diagram of collision detection using a convex scanning volume provided in Example 1.

[0075] Figure 6 This is a schematic diagram of the process of expanding new nodes using the adaptive sampling method in the random tree provided in Example 1.

[0076] Figure 7 This is a schematic diagram of a path planning device for an adaptive sampling rope-driven parallel handling robot provided in Example 2.

[0077] The present invention is further described in detail below with reference to the accompanying drawings and specific embodiments. DETAILED DESCRIPTION

[0078] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention. Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the invention for which protection is sought, but merely represents selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0079] Example 1

[0080] Embodiment 1 of the present invention provides a path planning method for an adaptive sampling rope-driven parallel transport robot, which can be implemented by a path planning device for an adaptive sampling rope-driven parallel transport robot (hereinafter referred to as the path planning device), and in particular, executed by one or more processors in the path planning device.

[0081] Embodiment 1 of the present invention provides a path planning method for an adaptive sampling rope-driven parallel transport robot, which can be implemented by a path planning device for an adaptive sampling rope-driven parallel transport robot (hereinafter referred to as the path planning device), and in particular, executed by one or more processors in the path planning device.

[0082] In this embodiment, the path planning device can be an electronic device equipped with a processor, which has a computer program for the path planning method of the adaptive sampling rope-driven parallel transport robot and can be executed, such as a computer, a smart phone, a smart tablet, a workstation, etc., which is not limited here.

[0083] like Figure 1 As shown, a path planning method for an adaptive sampling rope-driven parallel handling robot includes steps S1 to S7.

[0084] S1, obtain the starting point, target end point, environmental information, the initial state of the robot, and the workspace, and determine the robot's search space; wherein the robot's initial position is the same as the starting point; the environmental information includes the current position of the obstacle.

[0085] like Figure 2 The simplified diagram of a rope-driven parallel transport robot shown in Figure 1 first requires a clear understanding of the rope-driven parallel transport robot's workspace and obstacle distribution. Image processing algorithms can be used to automatically identify the position of objects using visual sensors (such as cameras), thereby determining the starting point (the initial position of the target object), the target end point (the target placement position), and the location of obstacles. The robot's initial state includes the robot's position (including the position of the end effector). In this embodiment, the robot's position is set to be the same as the target object's initial position, and is represented using a Cartesian coordinate system.

[0086] The planning path of the present invention includes a path planning process from the current position of the robot (ie, the initial position of the target object) to the target placement position.

[0087] The Cartesian coordinate system is a geometric positioning system based on three mutually perpendicular axes that intersect at a point. It can conveniently represent any point in three-dimensional space. Using the Cartesian coordinate system, a robot can accurately determine the position of an object and the posture of its end effector (such as a gripper).

[0088] The search space for path planning is defined by combining the motion range of the robot's end effector (i.e., workspace) and the obstacle space.

[0089] Let x∈R d , represents the d-dimensional state space of the rope-driven parallel transport robot, and R represents a real number;

[0090] Obstacle space x obs , Indicates the state where the robot collides with an obstacle;

[0091] Free Spacex free , x free =x\x obs A set of states that represent a robot that does not collide with obstacles.

[0092] The state of the robot at time t is x rob(t) ∈x.

[0093] In this embodiment, the dimensional state space of the rope-driven parallel transfer robot refers to the set describing all possible states of the robot. These states include dynamic parameters such as the robot's position, velocity, and acceleration, as well as static parameters such as the pose of the end effector. The size of the dimensional state space depends on the robot's complexity and the number of degrees of freedom.

[0094] In this embodiment, obstacle space refers to the set of states in which the robot can collide with obstacles. During robot motion planning, the robot must avoid entering obstacle space to prevent collisions. Free space, on the other hand, refers to the set of states in which collisions do not occur and is the region in which the robot can move normally. Using a sound path planning algorithm, the robot can safely move to its target destination in free space.

[0095] like Figure 2 The simplified diagram of the rope-driven parallel transport robot shown in the figure, A1 to A4 are fixed rope anchor points on the four end effectors of the rope-driven parallel transport robot, and B1 to B4 respectively represent the ends of the ropes connecting the end effectors to the robot body at the robot body, that is, the movable rope anchor points.

[0096] In this embodiment, the robot's position is described by the position vector of the end effector. The workspace of the cable-driven parallel transfer robot is defined as the set of feasible end effector poses, which includes the position and orientation of the end effector. This set includes all possible positions and poses that the robot's end effector can reach without colliding with other objects. The size and shape of the workspace depend on factors such as the robot's structural parameters, drive method, and environmental constraints.

[0097] Apply spatial constraints to the robot's workspace w. The spatial constraints imposed by the workspace can be expressed as:

[0098]

[0099] Among them, t f represents the total time; t represents the moment variable; x robIndicates the state of the robot; x rob(t) represents the state of the robot at time t; w represents the robot's workspace.

[0100] According to the robot's workspace w and obstacle space x obs , the robot’s search space can be determined.

[0101] S2, according to the position of the workspace and the current obstacle, check the initial state of the robot, and create a random tree with the starting point and the target end point as the root node respectively in a feasible initial state.

[0102] In this embodiment, the first requirement for a feasible trajectory is to ensure that the robot's state at the current time t belongs to the free space x free The range of x rob(t) ∈x free ,If it exceeds the workspace boundary, the robot cannot operate normally.

[0103] Then, consider whether the minimum distance between the robot and the obstacle is greater than the set safety threshold L, which is expressed as:

[0104]

[0105] Among them, C i (t), O i (t) represents the two points closest to the robot's end effector and the obstacle at time t, i = 0, 1, ... m; m represents the obstacle space x obs The center of the convex body. The robot's initial position overlaps with or is too close to a known obstacle, which may cause a collision or restricted operation.

[0106] Then the target end point x goal Need to meet:

[0107] x goal ={x∈x free |||C i (r)-O i (t)|| <r g};

[0108] Among them, r g Indicates the target end point x goal The distance parameter.

[0109] like Figure 3 As shown, when judging the minimum distance between the robot and the obstacle, the obstacle space x obs and the robot's end effector space x rob Modeled as a convex body (i.e., an axis-aligned bounding box), each convex body is defined by a set of vertices, then the obstacle space x obsThe convex vertex set M is expressed as:

[0110] M={x obs + m d1,...,x obs + m d i ,...,x obs + m d8};

[0111] in, m d i ,i=1,...8, represents the position vector of the i-th vertex in the obstacle space at the center m of the bounding box in the m-xyz coordinate system.

[0112] The robot's end effector space x rob The convex vertex set R′ is expressed as:

[0113] R′={x rob + r b1,...,x rob + r b j ,...,x rob + r b8};

[0114] in, r b j , j = 1, ... 8, represents the position vector of the j-th vertex of the end effector of the rope-driven parallel transfer robot at the center r of the bounding box in the r-xyz coordinate system.

[0115] The rope is a convex body defined by two extreme vertices, and the i-th rope is defined as:

[0116] L i ={a i , x rob + r b i};

[0117] Among them, a i Represents the vertex coordinates of the i-th rope.

[0118] See Figure 3 , m and r represent the obstacle space x obs Convex body and robot end effector space x rob The center of the convex body. m and C m Represent points in the obstacle space and the robot's end-effector space, respectively.

[0119] The vertex information of the convex body is extracted to form the corresponding vertex set. The GJK algorithm is used to detect the distance between the end effector and the rope of the rope-driven parallel transport robot and the obstacle in real time. That is, the Minkowski difference of the two convex body vertex sets is calculated, and the closest point in the Minkowski difference is found, which is the minimum distance between the robot and the obstacle.

[0120] In this embodiment, the GJK (Gilbert-Johnson-Keerthi) algorithm is an efficient algorithm for calculating the minimum distance between two convex bodies. It uses the concept of Minkowski difference sets in an iterative manner to quickly determine whether two geometric bodies intersect and calculate the minimum distance between them.

[0121] If the initial state satisfies the free space range and the safety threshold L, indicating that the current position is within the workspace of the rope-driven parallel transport robot and the distance to the obstacle is within the safe range, random trees T1 and T2 are created with the robot position in the initial state (i.e., the starting point) and the target end point as the root nodes, respectively, and attempts are made to expand from the existing nodes (starting from the root node for the first time) to the newly sampled nodes.

[0122] S3, based on the current state of the random tree, adopts a bidirectional adaptive sampling strategy to search and sample in their respective search spaces, continuously and alternately expanding the new nodes of the two random trees to find the next child node in the feasible path.

[0123] Traditional RRT algorithms (path planning algorithms) typically sample uniformly in free space, but this approach can be inefficient when approaching a target or in areas with dense obstacles. During random tree expansion sampling, an embodiment of the present invention proposes a bidirectional adaptive sampling and growth method for random sampling. Two random trees, T1 and T2, perform search sampling in their respective search spaces, improving the efficiency of random tree expansion, especially in complex environments. This method dynamically adjusts sampling and node expansion strategies based on the current state of the random tree and the target endpoint location, thereby finding a feasible path more quickly.

[0124] When performing adaptive sampling, if a path from the starting point to the target destination has been found, the algorithm will not perform uniform random sampling in the entire workspace, but will sample in an elliptical area with the current path as the core to further optimize the current path and make it shorter. The focus of the ellipse is the starting point x of the path. start and the target endpoint x goal The major axis of the ellipse is set according to the current path length (e.g., half the path length). This method can be used to further optimize the path found, making it shorter and more direct.

[0125] If no path to the target destination is found, the weights between sampling strategies are adaptively adjusted based on the node's expansion failure rate. Sampling strategies include target bias (focused on the target) and random sampling (focused on random sampling). Other sampling strategies can also be used as needed.

[0126] When the node expansion failure rate is less than the set threshold, it indicates that the current random tree is growing smoothly and encountering fewer obstacles. Therefore, the target bias strategy tends to be selected for expansion toward the target. Therefore, the target bias strategy weight α is increased, and sampling is performed toward the target. Otherwise, it indicates that the current random tree is encountering many obstacles. Therefore, the random sampling strategy weight is increased, and the search range is expanded for sampling to find a path around the obstacles. The node expansion failure rate is the ratio of the number of node expansion failures to the total number of node expansions.

[0127] The formula for adaptive weight sampling is:

[0128] P sample =α×P goal +(1-α)×P random ;

[0129] Among them, P sample is the sampling node position; α is the weight of the target bias strategy, 0≤α≤1; P goal is the position of the target direction; P random is a random position in free space.

[0130] like Figure 4 As shown, during the sampling process, after adaptive sampling generates a random node, it automatically searches for the node x in the random tree T1. rand The nearest node x nearest1 ; Node x nearest1 To a random node x rand Expand a step in the direction of , generating a new node x new1 .

[0131] New node x new1 Use its neighbor nodes to optimize the random tree T1 and select the neighbor node with the smallest path cost as x new1 The best parent node of the new node x is then added with the minimum cost. new1 Add child nodes.

[0132] Generate a new node x in the random tree T1 new1 After that, swap the current operation tree and try to connect the two random trees. j will find the nearest child node x in the random tree T2. nearest2 , along x nearest2 To x new1 Expand one step in the direction to generate a new node x new2. New node x new2 The random tree T2 is optimized by using its neighbor nodes. First, the path with the smallest cost is found by finding the best parent node, and then the path to the new node x is optimized with the smallest cost. new2 Add child nodes.

[0133] After generating a new node, it is necessary to determine whether the new node is the next child node in a feasible path. The GJK algorithm is used for fast collision detection, as it requires a sufficient number of alternative trajectories within a fixed interval. First, the generated new node is determined to be within the robot's workspace. If not, the new node is discarded and resampled and expanded. Otherwise, the minimum distance between the rope-driven parallel transport robot and the obstacle is checked against a safety threshold. If the distance is greater than the safety threshold, the path between the current node and its neighboring nodes is safe and feasible. If it is less than the safety threshold, the path from the current node's neighboring nodes to the current node is further estimated to determine whether there will be a collision between the obstacle and the path.

[0134] To simplify collision estimation, it is assumed that the motion of the rope-driven parallel transport robot is linearly driven, that is, the motion of the robot in any time period is approximated as linear motion. The motion of the rope-driven parallel transport robot involves the position change of the end effector and the motion of the rope. Figure 5 (a) with Figure 5 As can be seen in (b), the movable rope anchor point B i With fixed rope anchor point A i The motion area in space is triangular, x rob(t＝i) 、x rob(t＝i+1) This represents the state of the robot's end effector moving from the current time i to the next time i+1. The GJK algorithm is used to detect whether the rope-driven robot's end effector and rope will collide with obstacles in the spatial motion area. If not, the path between the current node and its neighboring nodes is safe and feasible, and the new node becomes the next child node. Otherwise, the new node is discarded and the sample is resampled.

[0135] S4, traverse each neighbor node within the range of the current child node found in turn, calculate the path cost from the root node of the current expanded random tree to the current child node, and select the neighbor node with the smallest path cost as the best parent node of the current child node.

[0136] In this embodiment, in order to find a better parent node and tree structure, reduce path cost and improve path quality, the new node and its neighboring nodes are optimized.

[0137] Traverse the new node x in sequence new For each adjacent node within the domain (i.e. the current child node found), the domain radius R is set to:

[0138]

[0139] Among them, R η >0 is a pre-constant, d′ represents the problem dimension, and n is the number of nodes.

[0140] Then, the path length from the root node to the current child node, i.e., the path cost, is calculated, and the neighbor node with the smallest path cost (i.e., the shortest path) is selected as the best parent node.

[0141] like Figure 6 (a) shows the random tree T1 before adaptive sampling. Figure 6 As shown in (b), an adaptive sampling method is used to generate a new node x 12 Then, the random tree T1 is optimized using its adjacent nodes {x5, x6, x9}. 12 The original path is {x0, x2, x5, x6, x 12}. 12 After the parent node is replaced by the neighbor node x5, a path with a lower path cost is found {x0, x2, x5, x 12}, so x 12 The path cost is reduced from 4 to 3. Find the new node x 12 After finding the best parent node of x, traverse the neighbor nodes again and move to x with the minimum cost. 12 Add child node x9.

[0142] S5, after each new node is generated, the current random tree is exchanged and two random trees are tried to be connected until the two random trees have the same node or the distance between the generated new nodes is less than the set threshold. The root nodes of the two random trees are used as the starting and end points, and the child nodes with the smallest path cost are searched to generate a heuristic trajectory to complete the construction of the random tree.

[0143] In this step, each time a random tree generates a new node, it will try to connect it with another random tree. If the new node expansion fails or the connection attempt between random trees T1 and T2 fails, the current random tree structure will be retained and the next iteration will generate a new random sampling point x. rand , and continue to expand random trees T1 and T2. If the two random trees search for the same node or the distance between the two new nodes generated by random trees T1 and T2 is less than the threshold, it means that the two random trees are successfully connected, and the algorithm will stop searching.

[0144] In this step, if the two random trees are successfully connected, starting from the root nodes of the two random trees, a bidirectional search is performed to generate a heuristic trajectory for the child nodes with the smallest path cost. The total cost function is constructed based on the heuristic function combined with the path cost, and the optimal branch in the path is selected by calculating the value of the total cost function. The calculation formula of the total cost function is:

[0145] f i =cost(x i )-h i ;

[0146] h i =||x i -x goal ||;

[0147] Among them, x i represents the current node, f i Represents the total cost function of the current node; cost(x i ) represents the path length of the current node, that is, the path from the root node to the current node x i The path cost of h i Represents the current node x i To the target end point x goal expected costs.

[0148] By comparing f i The optimal branch is selected by using the value. Figure 6 As shown in (c), after the random tree is constructed, starting from the root node of the tree, a heuristic trajectory {x0, x2, x5, x6, x7} is generated by continuously searching for child nodes with smaller costs. 11}.

[0149] If no heuristic trajectory is found, the rope-driven parallel transport robot stops on the spot to prevent danger until a new feasible trajectory is generated.

[0150] S6, according to the generated heuristic trajectory, control the robot to move from the current node along the trajectory to the next child node.

[0151] like Figure 6 As shown in (c), after the trajectory is generated, the rope-driven parallel transport robot is set at the root node x0, and the robot moves to the next child node x2. At the same time, the root of the planned trajectory is changed to x2, and the nodes with timestamps earlier than x0 and their child nodes {x1, x4, x8, x 10}, keeping the remaining tree.

[0152] S7, updates the robot's state and environment information, takes the robot's current state node as the root node of the new random tree, resamples and expands new nodes to search for a better path until the robot reaches the target end point.

[0153] In this step, the robot and environment states are updated at fixed time steps or when an obstacle is detected, and resampling and path planning are performed. The control rope drives the end effector of the parallel transport robot to move to the next direct node along the generated trajectory.

[0154] like Figure 6 As shown in (d), when generating a heuristic trajectory, useless branch nodes are deleted. When further obstacles are detected, the states of the robot and the environment are updated. The node x2 of the robot's current state is used as the root node of the new random tree. The next round of path planning continues, and new nodes are resampled and expanded to search for a better path until the robot reaches the target destination.

[0155] In summary, compared with the prior art, the present invention has the following beneficial effects:

[0156] During path planning, the present invention constructs a bidirectional random tree to generate a path from a starting point to a target destination. It then adaptively adjusts the weights of a growth strategy (including a target bias strategy and a random sampling strategy) to perform bidirectional sampling and expansion of new nodes, finding the next child node in a feasible path and improving the efficiency of random tree expansion. This method dynamically adjusts the sampling and expansion strategy based on the current state of the random tree and the target destination, and, combined with the path cost, allows for faster discovery of a safe and feasible path.

[0157] After generating the heuristic trajectory of the current position, the present invention deletes useless branch nodes, thereby reducing the memory of the algorithm and the computational complexity.

[0158] This invention combines updated robot status with environmental information to resample and plan paths, finding a more optimal path to cope with changing environments and mitigate risks. This invention is suitable for the cargo handling needs of rope-driven parallel transport robots. It fully considers their motion characteristics and operating conditions, and plans an optimal, safe, and feasible path for the robot to transport cargo in complex working environments.

[0159] Example 2

[0160] like Figure 7 As shown, the second embodiment of the present invention further provides a path planning device for an adaptive sampling rope-driven parallel transport robot, comprising:

[0161] An acquisition unit, configured to acquire a starting point, a target end point, environmental information, an initial state of the robot, and a workspace, and determine a search space for the robot; wherein the initial position of the robot is the same as the starting point; and the environmental information includes the current position of obstacles;

[0162] a random tree creation unit, configured to check an initial state of the robot according to the position of the workspace and the current obstacle, and create a random tree with the starting point and the target end point as root nodes respectively in a feasible initial state;

[0163] The bidirectional sampling unit is used to search and sample in each search space according to the current state of the random tree using a bidirectional adaptive sampling strategy to find the next child node in the feasible path;

[0164] The path cost calculation unit is used to traverse each neighbor node within the range of the current child node, calculate the path cost from the root node of the current extended random tree to the current child node, and select the neighbor node with the smallest path cost as the best parent node of the current child node;

[0165] The heuristic trajectory generation unit is used to alternately expand the new nodes of the two random trees and try to connect the two random trees after each new node is generated until the two random trees have the same node or the distance between the generated new nodes is less than the set threshold. With the root nodes of the two random trees as the starting and end points, the unit searches for the child nodes with the smallest path cost to generate a heuristic trajectory and complete the construction of the random trees.

[0166] The control unit is used to control the robot to move from the current node along the trajectory to the next child node according to the generated heuristic trajectory;

[0167] The update unit is used to update the robot's state and environment information, taking the robot's current state node as the root node of the new random tree, resampling and expanding new nodes to search for a better path until the robot reaches the target end point.

[0168] Example 3

[0169] The third embodiment of the present invention also provides a path planning device for an adaptive sampling rope-driven parallel transport robot, which includes a memory and a processor. The memory stores a computer program, and the computer program can be executed by the processor to implement the path planning method for the adaptive sampling rope-driven parallel transport robot as described above.

[0170] Example 4

[0171] The fourth embodiment of the present invention also provides a computer-readable storage medium, which stores computer-readable instructions. When the computer-readable instructions are executed by the processor of the device where the computer-readable storage medium is located, the path planning method of the adaptive sampling rope-driven parallel transport robot as described above is implemented.

[0172] In the several embodiments provided in the embodiments of the present invention, it should be understood that the disclosed devices and methods can also be implemented in other ways. The device and method embodiments described above are merely illustrative. For example, the flowcharts in the accompanying drawings show the possible architectures, functions, and operations of the devices, methods, and computer program products according to multiple embodiments of the present invention. In this regard, each box in the flowchart or block diagram can represent a module, program segment, or part of a code, which contains one or more executable instructions for implementing the specified logical functions. It should also be noted that in some alternative implementations, the functions marked in the boxes can also occur in an order different from that marked in the drawings. For example, two consecutive boxes can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flowchart, as well as the combination of boxes in the block diagram and / or flowchart, can be implemented using a dedicated hardware-based system that performs the specified functions or actions, or can be implemented using a combination of dedicated hardware and computer instructions.

[0173] In addition, the functional modules in the various embodiments of the present invention may be integrated together to form an independent part, or each module may exist independently, or two or more modules may be integrated to form an independent part.

[0174] If the functions are implemented in the form of software function modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, electronic device, or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes various media that can store program code, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk. It should be noted that, in this article, the terms "include", "comprising" or any other variant thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device that includes a series of elements includes not only those elements, but also other elements that are not explicitly listed, or also includes elements inherent to such a process, method, article or device. Without further constraints, an element defined by the phrase "comprises a..." does not preclude the existence of additional identical elements in the process, method, article or apparatus that includes the element.

[0175] The terms used in the embodiments of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. The singular forms "a", "an", "the" and "the" used in the embodiments of the present invention and the appended claims are also intended to include plural forms unless the context clearly indicates otherwise.

[0176] It should be understood that the term "and / or" as used herein is merely a description of the relationship between associated objects, indicating that three possible relationships exist. For example, "A and / or B" can represent: A exists alone, A and B exist simultaneously, or B exists alone. Furthermore, the character " / " in this document generally indicates that the associated objects are in an "or" relationship.

[0177] The word "if," as used herein, may be interpreted as "at the time of" or "when" or "in response to determining" or "in response to detecting," depending on the context. Similarly, the phrases "if it is determined" or "if (stated condition or event) is detected" may be interpreted as "when it is determined" or "in response to the determination" or "when detecting (stated condition or event)" or "in response to detecting (stated condition or event)," depending on the context.

[0178] The "first" and "second" mentioned in the embodiments are merely used to distinguish similar objects and do not represent a specific ordering of the objects. It is understood that the specific order or precedence of "first" and "second" can be interchanged where appropriate. It should be understood that the objects distinguished by "first" and "second" can be interchanged where appropriate, so that the embodiments described herein can be implemented in an order other than that illustrated or described herein.

[0179] The foregoing description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Those skilled in the art will readily appreciate that various modifications and variations of the present invention are possible. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention are intended to be within the scope of protection of the present invention.

Claims

1. A path planning method for an adaptive sampling rope driven parallel handling robot, characterized in that: include: S1, obtaining the starting point, target end point, environmental information, the initial state of the robot, and the workspace, and determining the robot's search space; wherein the robot's initial position is the same as the starting point; the environmental information includes the current location of obstacles; S2, checking the initial state of the robot according to the position of the workspace and the current obstacle, and creating a random tree with the starting point and the target end point as root nodes respectively in a feasible initial state; S3, based on the current state of the random tree, a bidirectional adaptive sampling strategy is used to search and sample in each search space to find the next child node in the feasible path; S4, traverse each neighbor node within the range of the current child node found in turn, calculate the path cost from the root node of the current expanded random tree to the current child node, and select the neighbor node with the smallest path cost as the best parent node of the current child node; S5, after each new node is generated, the new nodes of the two random trees are continuously expanded alternately and the two random trees are tried to be connected until the two random trees have the same node or the distance between the generated new nodes is less than the set threshold. With the root nodes of the two random trees as the starting and end points, the child nodes with the smallest path cost are searched to generate a heuristic trajectory to complete the construction of the random tree; S6, according to the generated heuristic trajectory, control the robot to move from the current node along the trajectory to the next child node; S7, updates the robot's state and environment information, takes the robot's current state node as the root node of the new random tree, resamples and expands new nodes to search for a better path until the robot reaches the target end point.

2. A path planning method for an adaptive sampling rope driven parallel handling robot according to claim 1, characterized in that ,The search space of the robot is determined according to the robot’s workspace and the obstacle space; ,the workspace of the robot is a set of feasible poses of its end ,effector. The pose includes the position and orientation of the end ,effector, and the position is expressed in a Cartesian coordinate system; Apply spatial constraints to the robot's workspace. The spatial constraints are expressed as: x rob(t) ∈x,x∈R d ; Among them, t f represents the total time; t represents the moment variable; x rob represents the robot's end effector space; x rob(t) represents the state of the robot at time t; w represents the robot's workspace; x represents the robot's d-dimensional state space; R represents a real number; Obstacle space x obs , Indicates the state where the robot collides with an obstacle; Free Spacex free , x free =x\x ons A set of states that represent a robot that does not collide with obstacles.

3. The path planning method for an adaptive sampling rope driven parallel handling robot according to claim 2, characterized in that ,When the robot meets the following conditions, it is a feasible initial state of the robot: The initial state of the robot satisfies the spatial constraints of the workspace; The state of the robot at the current time t belongs to the free space x free The range of x rob(t) ∈x free ; Calculate the minimum distance between the robot and the obstacle, which is within the safety threshold range, that is: Among them, C i (t), O i (t) represents the two points closest to the robot's end effector and the obstacle at time t; L represents the set safety threshold.

4. A path planning method for an adaptive sampling rope driven parallel handling robot according to claim 3, characterized in that When calculating the minimum distance between the robot and the obstacle, the robot's end effector and the obstacle are modeled as convex bodies, and the vertex information of the convex body is extracted to form the corresponding vertex set; and the GJK algorithm is used to calculate the minimum distance between the robot and the obstacle, that is, to calculate the Minkowski difference set of the two convex body vertex sets, and find the closest point in the Minkowski difference set, which is the minimum distance between the robot and the obstacle.

5. The path planning method for an adaptive sampling rope driven parallel handling robot according to claim 1, characterized in that ,The specific adaptive sampling strategy is as follows: If a path to the target destination has been found, sampling is performed within an elliptical area with the current path as the core to optimize the current path; wherein the focus of the ellipse is determined based on the target starting point and the target destination, and the major axis of the ellipse is set based on the length of the current path; If no path to the target destination is found, the weight between the target bias strategy and the random sampling strategy is adaptively adjusted according to the node expansion failure rate; When the node expansion failure rate is less than the set threshold, the weight of the target bias strategy is increased, and the sampling is expanded in the target direction; otherwise, the weight of the random sampling strategy is increased, and the search range is expanded for sampling; the node expansion failure rate is the ratio of the number of node expansion failures to the total number of expansions; The formula for adaptive sampling is: P sample =α×P goal +(1-a)×P random ; Among them, P sample is the sampling node position; α is the weight of the target bias strategy, 0≤α≤1; P goal is the position of the target direction; P random is a random position in free space; A bidirectional adaptive sampling strategy is used to sample a random node x in the robot’s search space. rand After that, get the random node x in the current random tree T1 rand The nearest child node x nearest1 , and along x nearest1 To a random node x rand Expand one step in the direction to generate a new node x new1 ; New node x new1 Use its neighbor nodes to optimize the random tree T1 and select the neighbor node with the smallest path cost as x new1 The best parent node of Generate a new node x new1 After that, swap the current operation tree, try to connect the two random trees, and find the closest child node x in the random tree T2. nearest2 , along x nearest2 To x new1 Expand one step in the direction to generate a new node x new2 ; New node x new2 Use its neighbor nodes to optimize the random tree T2 and select the neighbor node with the smallest path cost as x new2 The best parent node of If the new node expansion fails or the connection attempt between two random trees fails, the current random tree structure is retained and the next iteration is started to generate a new random sampling point x rand , and continuously expand random trees T1 and T2 alternately.

6. The path planning method for an adaptive sampling rope driven parallel handling robot according to claim 1, characterized in that ,When judging whether the new node is the next child node in the feasible path, the judgment method is: The GJK algorithm is used to detect whether the generated new node is in the robot's workspace. If not, the new node is discarded and resampled and expanded; Otherwise, determine whether the minimum distance between the robot and the obstacle at the new node is greater than the safety threshold; If it is greater, the new node is the next child node; Otherwise, further estimate whether there is a collision between the path from the new node's neighbor node to the new node and the obstacle; if there is no collision, it is a safe and feasible path, and the new node is the next child node; Otherwise, discard the new node and resample.

7. The path planning method for an adaptive sampling rope driven parallel handling robot according to claim 1, characterized in that ,When traversing each neighbor node within the domain of the current child node of the current expanded random tree, the domain radius R is set to: Among them, R η >0 is a pre-constant, d′ represents the problem dimension, and n is the number of nodes.

8. The path planning method for an adaptive sampling rope driven parallel handling robot according to claim 1, characterized in that , taking the root nodes of two random trees as the starting point and the end point, searching for the child nodes with the smallest path cost to generate a heuristic trajectory, constructing a total cost function based on the heuristic function combined with the path cost, and selecting the optimal branch in the path by calculating the total cost value; wherein, the calculation formula of the total cost function is: f i =cost(x i )-h i ; h i =||x i -x goal ||; Among them, x i represents the current node, f i Represents the total cost function of the current node; cost(x i ) represents the path length of the current node, that is, the path from the root node to the current node x i The path cost of h i Represents the current node x i To the target end point x goal expected costs.

9. A path planning method for an adaptive sampling rope driven parallel handling robot according to claim 8, characterized in that ,It also includes: if no feasible heuristic trajectory is generated, the robot stops in place to prevent danger until a new feasible trajectory is generated; when the robot moves from the current node to the next child node, the node and its child nodes with a timestamp earlier than the current node are deleted to optimize the path.

10. A path planning device for an adaptive sampling rope driven parallel handling robot, characterized in that: include: An acquisition unit, configured to acquire a starting point, a target end point, environmental information, an initial state of the robot, and a workspace, and determine a search space for the robot; wherein the initial position of the robot is the same as the starting point; and the environmental information includes the current position of obstacles; a random tree creation unit, configured to check an initial state of the robot according to the position of the workspace and the current obstacle, and create a random tree with the starting point and the target end point as root nodes respectively in a feasible initial state; The bidirectional sampling unit is used to search and sample in each search space according to the current state of the random tree using a bidirectional adaptive sampling strategy to find the next child node in the feasible path; The path cost calculation unit is used to traverse each neighbor node within the range of the current child node, calculate the path cost from the root node of the current extended random tree to the current child node, and select the neighbor node with the smallest path cost as the best parent node of the current child node; The heuristic trajectory generation unit is used to alternately expand the new nodes of the two random trees and try to connect the two random trees after each new node is generated until the two random trees have the same node or the distance between the generated new nodes is less than the set threshold. With the root nodes of the two random trees as the starting and end points, the unit searches for the child nodes with the smallest path cost to generate a heuristic trajectory and complete the construction of the random trees. The control unit is used to control the robot to move from the current node along the trajectory to the next child node according to the generated heuristic trajectory; The update unit is used to update the robot's state and environment information, taking the robot's current state node as the root node of the new random tree, resampling and expanding new nodes to search for a better path until the robot reaches the target end point.

Images